Research On Visual Servo High-Precision Positioning Control For LED Chip Sorting

The accuracy of visual positioning in chip sorting machine directly affects the performance of sorting process and chip product quality.Most of the chip vision positioning adopts the control method of scanning first and then positioning,i.e.,all chips on the wafer tray are scanned first and their position information is recorded,and then chip alignment is performed.At present,the visual positioning system applied in the sorting machine mainly has the following problems.Firstly,the error sources affecting the positioning accuracy of the system are complex,which makes it difficult to target the system performance;secondly,there are dynamic disturbances such as blue film deformation during the chip movement,and the system cannot detect the chip position changes in real time;in addition,the calibration parameters of the chip sorting process change,which has an impact on the positioning accuracy of the system.The workflow of LED chip sorter was introduced,and the effects of wafer disc installation angle,image recognition accuracy,blue membrane deformation and rotating arm vibration on system positioning accuracy were analyzed,and corresponding compensation methods were proposed respectively.Based on the positioning requirements of the chip sorter,the LED chip visual positioning experiment platform was built,and the system operation software and human-computer interaction interface were developed.To address the complex problem of error sources affecting system performance,an error analysis and compensation method based on Monte Carlo simulation was proposed.Firstly,the error sources affecting the positioning accuracy of the system were analyzed and measured,the error model of the visual positioning system was established,and the simulated chip positioning errors were obtained by Monte Carlo method,and the two sets of simulated and experimentally measured positioning error data were tested for independent identical distribution to verify the validity of the error model.Based on the error model of the system,the error transfer theory was used to calculate the influence factor of each error source on the positioning accuracy of the system and to determine the main error sources.Based on the error analysis results,an error compensation strategy was proposed to give priority to compensate the main error sources,and the optimality of the compensation method was judged by Monte Carlo simulation method.Finally,the effectiveness of the error analysis method and the error compensation method were verified through experiments.The visual servo positioning control method based on segmented continuous observer was proposed for the blue membrane deformation problem during the motion of LED chips.Firstly,the timing model of the system was analyzed,the dynamics model of the motion platform was established,the problem of delay and long sampling period of the visual signal was solved by the piecewise continuous observer,and the estimated real-time continuous visual signal was used as the position loop feedback to realize the full closed-loop control.The stability of the observer was demonstrated,and the effectiveness of the observer was verified by simulation results.The experimental results of LED chip positioning showed that the proposed method can compensate dynamic errors such as blue membrane deformation and achieve accurate fully closed-loop visual servo control.The adaptive recursive least squares uncalibrated visual servo positioning control method based on image pre-processing was proposed for the problem of calibration parameter variation during LED chip sorting.The kinematic model of the visual positioning control system was firstly established,and the recursive least squares method was used to estimate the calibration parameters of the system in real time,while the adaptive forgetting factor was introduced to increase the weight of new incoming data to accelerate the convergence of the algorithm.For the problem that the image signal with sampling delay affected the algorithm effect,the image pre-processing method was proposed.Robust Kalman filtering was used to compensate for the delayed image signal,and polynomial fitting was used to estimate the signal in non-sampling time,and adaptive fitting order was proposed to improve the accuracy of polynomial fitting.Finally,the effectiveness of the method was verified by comparing the estimation effect of the system calibration parameters before and after image pre-processing through simulation.The LED chip positioning experiments showed that the proposed uncalibrated visual servo positioning control method could guarantee the positioning accuracy of the system and meet the application requirements of LED chip sorting equipment when the calibration parameters were changed.